829
The Canadian Mineralogist Vol. 58, pp. 829-845 (2020) DOI: 10.3749/canmin.1900084
BERYLLIUM-SILICON DISORDER AND RARE EARTH CRYSTAL CHEMISTRY IN GADOLINITE FROM THE WHITE CLOUD PEGMATITE, COLORADO, USA
§,* † JULIEN M. ALLAZ ,JOSEPH R. SMYTH, RHIANA E. HENRY , AND CHARLES R. STERN Department of Geological Sciences, University of Colorado, 2200 Colorado Avenue, Boulder, Colorado 80309, USA
PHILIP PERSSON Department of Geology, Colorado School of Mines, 1516 Illinois Street, Golden, Colorado 80401, USA
JOY J. MA AND MARKUS B. RASCHKE Department of Physics, Department of Chemistry, and JILA, University of Colorado, 2000 Colorado Avenue, Boulder, Colorado 80302, USA
ABSTRACT
Gadolinite, REE2FeBe2Si2O10, is a monoclinic orthosilicate member of the gadolinite supergroup of minerals and occurs in beryllium and rare earth element (REE) bearing granites, pegmatites, and some metamorphic rocks. Gadolinite from the White Cloud pegmatite, South Platte Pegmatite district, Colorado, USA, has been investigated and shows unusually variable REE compositions and distinct Be-Si disorder. Crystal structure and chemistry of two petrographically distinct gadolinite samples from this locality have been studied by electron microprobe chemical analysis, laser ablation inductively coupled plasma-mass spectrometry (LA-ICP-MS), single-crystal X-ray diffraction (XRD), and micro-Raman spectroscopy. Within these samples, the gadolinite was found to range from gadolinite-(Y) to gadolinite-(Ce). Regions of nearly full occupancy of Fe at the M site, and partial substitution of Si for Be at the Q tetrahedral site, as well as substitution of Be for Si at the T site were observed, with up to 15% vacancy at the Fe site and up to 15% disorder between Be and Si at distinct tetrahedral sites elsewhere. The layered nature of the crystal structure allows for large variation of the radius of the cation at the A site which contains the REE. This study shows that Be may substitute for Si and that Be may be more abundant in geochemical systems than previously assumed.
Keywords: gadolinite, rare earth elements, beryllium, structural disorder, single crystal X-ray diffraction, electron microprobe, Raman spectroscopy.
INTRODUCTION Be2Si2O10, three sub-species have been distinguished depending on the major REE: gadolinite-(Ce) The gadolinite supergroup is a group of structurally (Segalstad & Larsen 1978), gadolinite-(Y) related minerals with general chemical formula (Miyawaki et al. 1984), and gadolinite-(Nd) (Skodaˇ A2MQ2T2O8u2, with principal cations A ¼ Ca, REE; et al. 2016, 2018). Although usually considered an M ¼ Fe, Ca, A (vacancy); Q ¼ Be, B; T ¼ Si, P, As; orthosilicate, gadolinite has a layered crystal structure, and u ¼ O, OH, F, and with a monoclinic unit cell with with sheets of crystallographically distinct Be (Q) and space group P21/c (Bacˇ´ık et al. 2017 and references Si (T) tetrahedra forming four- and eight-membered 4þ 2þ therein). Of the specific species gadolinite REE2Fe rings of alternating Si and Be tetrahedra (Fig. 1).
§ Corresponding author e-mail address: [email protected] * Present address: ETH Zu¨rich, Department of Earth Sciences, Institute of Geochemistry and Petrology, Clausiusstrasse 25, 8092 Zu¨rich, Switzerland. † Present address: University of British Columbia, Department of Earth, Ocean, and Atmospheric Sciences, 2020 – 2207 Main Mall,Vancouver, BC, V6T 1Z4, Canada.
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 830 THE CANADIAN MINERALOGIST
FIG. 1. Crystal structure of gadolinite-(Y). (a) Top view of the (100) sheet. (b) Perpendicular to the (100) sheet. (c) TO4 and QO4 tetrahedra in alternating eight- and four-membered rings.
Between the sheets of tetrahedra, there is an interlayer solution with datolite, there is an additional coupled of REE3þ at eight-coordinated A sites and Fe2þ at substitution of Be2þ for B3þ and REE3þ for Ca2þ.As octahedrally coordinated M sites. There are four divalent Ca has a larger ionic radius than most distinct oxygen sites (O1–O4) plus a u anion site, all trivalent REE cations (with the exceptions of La3þ, occupied by O in gadolinite (Fig. 1a, b). All atoms are Ce3þ, and Pr3þ), the eight-membered tetrahedral ring at general positions in P2 /c so there are four of each 1 widens (Fig. 1c) so that the Si–Si short distance per unit cell, except M, which is at a special position at increases. Gadolinite subgroup minerals tend to have a the origin so there are two per unit cell. longer Si–Si long distance than the datolite subgroup Gadolinite commonly shows solid solution towards (Bacˇ´ık et al. 2014). The tetrahedral rings display a the isostructural hingganite, REE2ABe2Si2O8(OH,F)2 negative correlation between short and long Si–Si and datolite, Ca2AB2Si2O8(OH,F)2. In solid solution with hingganite or datolite, the octahedral M site of distances for gadolinite group minerals. The Si–Si long gadolinite becomes partially vacant, with charge distance is correlated with the occupancy of the T site, compensation via hydroxylation of the u site. In solid whereas the occupancies of the A and M sites are
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 Be-Si DISORDER AND REE CRYSTAL CHEMISTRY IN GADOLINITE 831
correlated with the Si–Si short distance (Bacˇ´ık et al. assemblage (sample Z2A) occurs as veins and pods 2014). up to ~2 m in maximum dimension composed of Gadolinite is commonly found in granite pegma- quartz, microcline, fluorite, thalenite-(Y),´ allanite- tites (e.g., Nilssen 1973, Pezzotta et al. 1999, Cernˇ y&´ (Ce), REE-carbonates, and subordinate euhedral to Ercit 2005, Bacˇ´ık et al. 2014). Despite its significant subhedral gadolinite. These two assemblages are presence in the White Cloud pegmatite, South Platte spatially separate but do occur in relatively close pegmatite district within the Pikes Peak granite in proximity (~10 m apart) to each other along the N/ Colorado, USA (Haynes 1965, Simmons & Heinrich NE perimeter of the core-intermediate zone contact 1980), there has been only limited characterization of of the White Cloud pegmatite. Much of the core and gadolinite from this locality. The goal of the present significant quantities of the REE zone have been study is to place gadolinite from this occurrence within mined, so the original extent and continuity of the the context of the recently refined spectrum of REE mineral-bearing zones is largely unknown gadolinite-group crystal chemistry (Bacˇ´ık et al. 2014, (Simmons & Heinrich 1980). 2017). Using a combination of electron microprobe (EMP) analysis, laser ablation inductively coupled SAMPLE DESCRIPTION plasma mass spectrometry (LA-ICP-MS), single crys- tal X-ray diffraction (XRD), and micro-Raman Two thin sections, SWC and Z2A, were used for spectroscopy, the Be-Si disorder in gadolinite is the EMP and micro-Raman spectroscopic analyses characterized and the crystallographic distortion due (Fig. 2). An additional thin section, WC1-A, was to compositional variation from light REE-rich prepared for additional petrographic studies and (LREE: La to Eu) to Y þ heavy REE-rich (HREE: complementary in situ analyses; this additional sample Gd to Lu) species is discussed. is an equivalent of sample Z2A. Gadolinite in SWC is a rounded centimeter-sized olive-green mass, associ- GEOLOGICAL SETTING ated with fluorite, feldspar, and quartz, with dark brown allanite-(Ce) inclusions. Simmons & Heinrich The White Cloud pegmatite is part of the South (1980) described a close association of gadolinite and Platte pegmatite district (Haynes 1965, Simmons & allanite-(Ce) and a dark reddish brown bastnasite-(Ce)¨ Heinrich 1980), which is hosted within the Protero- rim resulting from the alteration of REE-silicate zoic anorogenic ‘‘A-type’’ Pikes Peak granite (Smith minerals. The investigated samples do not display et al. 1999). This pegmatite district spans 80 km2 and such a bastnasite¨ rim, and only scarce REE-carbonates contains over 50 decameter-sized pegmatites concen- are found locally along the rim or within cracks. In trated in the north end of the batholith. The South sample Z2A, the gadolinite crystals are 1–4 mm across Platte pegmatites are well known for their local in a matrix of fluorite and are surrounded by allanite- concentrations of REE minerals, suggesting an REE (Ce), thalenite-(Y),´ and synchysite-(Y). All of the enrichment, and depletion in boron, as suggested by gadolinite crystals show an oscillatory zonation visible the absence of tourmaline. Many of these pegmatites in cross-polarized light (Fig. 2f, h). Sample Z2A is are concentrically zoned, with a border zone of more homogeneous, with less optical variation. graphic granite enriched in biotite, a K-feldspar-rich intermediate domain, and a quartz-rich core. The METHODS White Cloud pegmatite is ~100 3 40 m, and the Single crystal X-ray diffraction REE-rich phases are concentrated in pods and veins at the interface between the intermediate microcline- Crystal chips of ca.100lm diameter were extracted rich zone and the quartz core. This mineralization using a micro-drill from samples SWC and Z2A and consists of several REE-rich minerals including were prepared for single-crystal XRD. For both allanite-(Ce), gadolinite-(Ce,Y), thalenite-(Y),´ REE- samples, care was taken to extract a crystal chip carbonates, and fluocerite. Gadolinite appears to showing minimal compositional variation, as ascer- occur throughout all REE mineral-bearing zones of tained by EMP analyses. The XRD analyses were the White Cloud pegmatite and seems concentrated in conducted at the University of Colorado, Boulder, using two main assemblages. The first assemblage (sample aBrukerP4 four-circle diffractometer with an APEX II SWC) is most abundant and consists of a fluorite-rich CCD detector and a rotating Mo-anode generator. The mass up to ~3 m across composed of green to purple Mo-anode generator was operated at 50 kV and 250 mA anhedral fluorite and quartz, which hosts relatively with a graphite monochromator. Sample SWC was abundant disseminated anhedral gadolinite grains 2–5 analyzed through 2h of 708, and Z2A was analyzed mm in size (up to ~3 cm) and associated with REE through 2h of 808. Atom positions, anisotropic carbonates and minor allanite-(Ce). The second displacement parameters, and site occupancy were
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 832 THE CANADIAN MINERALOGIST
FIG. 2. Microphotographs of gadolinite minerals in samples SWC (a, b, e, f) and Z2A (c, d, g, h). PL: plane-polarized light, XPL: cross-polarized light. Aln ¼ allanite, Gad ¼ gadolinite, Fl ¼ fluorite, Qtz ¼ quartz. Note the oscillatory zoning clearly visible optically in (f) and (h).
refined for each specimen using SHELXL Version 1968, Azavant & Lichanot 1993). The A-site cation 2016/4 (Sheldrick 2015) in the WinGX-2014 software occupancies were refined using a Yb3þ scattering factor package (Farrugia 2014). Ionized-atom scattering because it has a high atomic number. The structural factors were used for refinements (Cromer & Mann occupancy appears low since the number of electrons
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 Be-Si DISORDER AND REE CRYSTAL CHEMISTRY IN GADOLINITE 833
detected for a mix of REE and Y is less than that for analysis at these conditions. Additional analytical pure Yb. The M-site occupancy was refined using the details and a list of standards used is given in Table Fe2þ scattering factor. Both Si and Be were allowed to 1. In addition to the analysis of thin sections, the crystal occupy the T and Q sites. The oxygen sites were fixed at chip of sample SWC was mounted on a glass slide after full occupancy (5 O, Z ¼ 2) and allowed to vary the XRD and polished for additional EMP analysis to spatially. Twinning instructions were included in the confirm the composition. Micro-Raman spectroscopy SHELXL refinement consistent with those of Ca´mara et was performed after EMPA and carbon removal on the al. (2008). The R-factor improved with the twin matrix selected regions of samples Z2A and SWC. applied to the refinement, although the twin fraction refined to less than one percent in both refinements. Laser ablation-inductively coupled plasma-mass spectrometry Raman spectroscopy Lithophile trace elements, Be, and B concentrations Micro-Raman spectroscopy and imaging of thin in gadolinite were measured for sample WC-1A with sections of SWC and Z2A were performed in the laser ablation sector field ICP-MS (LA-ICP-MS) at the Nano-Optics Laboratory, University of Colorado-Boul- Institute of Geochemistry and Petrology, ETH Z¨urich. der, using a specially designed homebuilt confocal Details of the analytical instruments are described in Raman microscope (based on an upright Olympus Guillong et al. (2014). Ablation times were 30 s for BX51 microscope, with an Olympus MPlanFL N 503/ individual measurements at energies of 2J/cm2 and 0.80NA BD objective, and a Princeton Instruments repetition rates of 4Hz, using a laser spot size of 19lm. Acton SpectraPro SP2500 spectrograph, with liquid NIST610 was used as a primary reference material and nitrogen-cooled CCD camera). The Raman spectra were GSD-1G was used as validation reference material collected in epi-illumination using 632.8 nm He-Ne laser with identical ablation conditions as the sample. The light, with 20 mW, focused onto the sample. Spectral EPMA measurements of Y obtained from the same resolution is 3.0 cm–1 (full width at half maximum) with spot were used for internal standardization. The results a 600 lines/mm grating and a confocal pinhole. Helium are given in Supplementary Table 21. and Ne lamp emission lines were used for calibration. Individual point spectra and arrays were acquired from RESULTS 50 to 4000 cm–1 with 50 s acquisition time per pixel. The arrays were acquired using a P-545.xR8S PInano XYZ X-ray diffraction Piezo scanner with 10 lm step size. The crystal structure refinement yielded a mono- Electron microprobe clinic unit cell with space group P21/c, with parameters displayed in Table 2. The cell parameters of SWC are The EMP analyses and backscattered electron (BSE) significantly larger than those of Z2A. Selected nearest- images were collected with a JEOL JXA-8230 neighbor distances are listed in Table 3. The u site is microprobe at the University of Colorado-Boulder. included in the nearest neighbor distances as an oxygen The instrument is equipped with five WDS spectrom- atom, based on the assumption of 10 oxygen atoms per eters, four of which are equipped with large-area formula unit. The refined atomic positions and 2 monochromators, and a 10 mm Thermo UltraDry displacement parameters are listed in Table 4. The EDS. The EDS detector was used to identify the refined occupancies of various sites are given in Table mineral assemblage and the major compositional 5. The small amount of water (see Raman below) was variations within individual crystals. In parallel, BSE neglected for the structure refinement. However, imaging of SWC and Z2A was used to identify zonation considering the presence of numerous heavy elements within the gadolinite and to select regions for in gadolinite, the presence or absence of minor amounts quantitative analyses in order to capture the composi- of H in the u site does not affect the refinement results. tional variability. All analyses were collected with an Significant positive residuals were observed near the A acceleration voltage of 15 keV, a 20 nm carbon coating, sites in both refinements. This is not unexpected for a and using the software Probe for EPMA.Forthelargest site with a large number of different chemical homogeneous domains, a 5 lm defocused beam at 50 substituents. Atom fragments were added to these nA current was used to prevent beam damage. For smaller domains, the current was decreased to 30 nA and the beam was defocused to 2 lm. All analyses were 1 Supplementary Data are available from the Depository of acquired using the Mean Atomic Number (MAN) Unpublished Data on the MAC website (http:// background correction (Donovan & Tingle 1996). No mineralogicalassociation.ca/), document ‘‘Gadolinite, beam damage was observed during the four-minute CM58, 19-00084’’.
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 834 THE CANADIAN MINERALOGIST
TABLE 1. ANALYTICAL CONDITIONS AND STANDARDS USED FOR EMP ANALYSES
† Element Crystal Standard Pos. (mm) tPeak (s) * DL Interferences
Y La TAP 1 YPO4 70.14 60 140–200 Si Si Ka TAP 1 Amelia Albite 77.45 30 65–85 Ho, Tm, Yb, Y Mg Ka TAP 1 Diopside 107.48 90 40–60 Tb, Eu, Nd, Dy Al Ka TAP 1 Almandine NY 90.63 90 35–55 Tm, Ho, Yb, Er
Th Ma PETL 2 ThO2 132.41 90 210–270 - U Mb PETL 2 UO2 118.88 90 170–220 Th Ca Ka PETL 2 Fluorapatite 107.48 90 25–35 Yb, Dy Pb Mb PETL 3 Pyromorphite 162.55 300 100–135 Ce
La La LIFL 4 LaPO4 185.38 40 265–350 Nd Ce La LIFL 4 CePO4 178.14 40 240–315 - Nd La LIFL 4 NdPO4 164.84 40 220–290 Ce Pr Lb LIFL 4 PrPO4 157.09 40 400–530 - Sm Lb LIFL 4 SmPO4 138.98 40 420–550 Tb, Er Tb La LIFL 4 TbPO4 137.42 40 245–325 Sm, Pr, Tm Eu Lb LIFL 4 EuPO4 133.58 40 430–570 Dy
Dy La LIFL 5 DyPO4 132.68 30 260–345 Eu, Yb, Sm Gd Lb LIFL 5 GdPO4 128.34 30 470–625 Ho Er La LIFL 5 ErPO4 124.01 40 240–315 Tb, Eu Yb La LIFL 5 YbPO4 116.14 40 265–355 Dy, Eu, Tb, Sm Ho Lb LIFL 5 HoPO4 114.44 40 475–625 Eu, Sm, Dy Tm La LIFL 5 TmPO4 119.97 40 255–340 Dy, Sm, Tb Fe Ka LIFL 5 Almandine NY 134.56 20 115–150 Dy, Eu, Nd, Sm
Lu La LIFL 5 LuPO4 112.47 40 285–380 Dy, Ho * Only a peak measurement was acquired, as the MAN background correction was used (Donovan & Tingle 1996). See Allaz et al. (2019) for a comparison of multipoint-background and MAN background acquisitions. † DL ¼ detection limit in ppm.
positions in the final stages of refinement, which However, the White Cloud gadolinite minerals are all significantly reduced the final R values. The XRD low in actinide contents (see below) and do not appear structural refinements converged to an R value of 0.049 to have sufficient radiation damage to affect the for SWC and 0.053 for Z2A. These are robust refinements. refinements for a chemically complex structure. Gibson & Ehlmann (1970) demonstrated that metamict Raman spectroscopy gadolinite crystals give poor diffraction patterns. Representative Raman spectra for samples SWC and Z2A are shown in Figure 3a–b. The bands between TABLE 2. X-RAY DIFFRACTION DATA AND 200 and 750 cm–1 are due to bending modes of Si–O REFINEMENT PARAMETERS FOR GADOLINITE and Be–O, stretching vibrations of REE–O and Fe–O, SAMPLES and other lattice vibrations (Skodaˇ et al. 2016). Both samples have their most prominent peak between 900 Parameter SWC Z2A and 914 cm–1 and a minor peak around 980 cm–1. a (A˚ ) 4.8201(2) 4.7770(9) These peaks are attributed to stretching vibrations of b (A˚ ) 7.6883(3) 7.5604(21) Be–O and Si–O in tetrahedral coordination (Skodaˇ et c (A˚ ) 10.1347(5) 10.0166(24) al. 2016), similar to phenakite (Be2SiO4). The weak b (8) 90.009(3) 90.402(12) spectral features in the 3000 to 3600 cm–1 region ˚ 3 Volume (A ) 375.58 361.75 correspond to different OH-stretching vibrations of 2h Max (8) 70.2 80.7 water. The underlying background in the 2900–4000 # Reflections 10761 9938 cm–1 region is due to fluorescence with considerable #Unique Reflections 1607 2158 spatial variability. R(int) 0.047 0.054 R 0.048 0.049 Raman hyperspectral imaging was performed on samples SWC and Z2A across regions with varying
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 Be-Si DISORDER AND REE CRYSTAL CHEMISTRY IN GADOLINITE 835
TABLE 3. SELECTED NEAREST NEIGHBOR heavy REE (HREE ¼ Y þ Gd to Lu) and light REE DISTANCES, POLYHEDRAL VOLUMES, QUADRATIC (LREE ¼ La to Eu) contents, as identified with EMP ELONGATION, ANGLE VARIANCE, AND ELECTRON analysis (BSE images in Fig. 3c, d of locations OCCUPANCIES FOR CATION SITES IN GADOLINITE indicated in Fig. 4c, d). The BSE images are correlated SAMPLES with images of values of Raman peak position m~ (top panel) and spectral linewidth C (middle panel) A site SWC Z2A obtained for the tetrahedral Be–O stretching vibration A–O1 (A˚ ) 2.385(5) 2.346(4) between 900 and 914 cm–1, from spectral fits to A–O1 (A˚ ) 2.388(5) 2.320(4) Lorentzian and Voigt line shapes (Fig. 3c–d). A ˚ A–O2 (A) 2.452(4) 2.411(3) distinct increase in linewidth correlated with spectral ˚ A–O3 (A) 2.730(5) 2.478(4) red-shift is observed and associated with the spatial ˚ A–O3 (A) 2.547(5) 2.694(4) variation in LREE versus HREE content, except in a A–O4 (A˚ ) 2.450(4) 2.413(4) certain region of sample SWC (Fig. 3 c, e) that seems A–u (A˚ ) 2.536(5) 2.384(4) structurally more disordered. A larger linewidth and A–u (A˚ ) 2.432(5) 2.496(4) ,A–O. (A˚ ) 2.490 2.443 more pronounced red-shift in SWC compared to Z2A Polyhedral Volume (A˚ 3) 26.3(4) 24.7(2) suggest greater disorder, which we attribute to the Electrons at Site 46.2(4) 43.6(3) increased Be–Si disorder as well as higher amounts of Th in sample SWC. M site SWC Z2A Electron microprobe analysis M–O2(2) (A˚ ) 2.324(5) 2.289(4) M–O4(2) (A˚ ) 2.227(5) 2.203(4) The variation of BSE intensity correlates with low- M–u(2) (A˚ ) 2.050(5) 2.049(4) density domains rich in YþHREE and brighter high- ,M–O. (A˚ ) 2.201 2.180 density domains rich in LREE (Fig. 4). The zoning is ˚ 3 Polyhedral Volume (A ) 13.2(2) 12.9(1) particularly prominent in one exceptionally large O. Q. E. 1.052 1.048 agglomerate of euhedral to subhedral gadolinite Angle Variance 162.2 154.8 crystals in sample WC1-A (Fig. 4a), with core Electrons at Site 21.1(3) 22.7(2) domains rich in LREE and rims richer in HREE. Some core domains appear patchy (Fig. 4d, e). Q site SWC Z2A Oscillatory zonation with alternation of Ce-rich and Q–O2 (A˚ ) 1.668(7) 1.648(6) Y-rich domains is present in all samples (Fig. 4a, c, e). Q–O3 (A˚ ) 1.679(7) 1.667(6) Rare REE-carbonates and Fe-oxides are present in ˚ Q–O4 (A) 1.648(8) 1.654(7) cracks in a few grains (Fig. 4b). The edges of some ˚ Q–O5 (A) 1.598(8) 1.594(6) grains show some minimal alteration (Fig. 4c). Sample ,Q–O. (A˚ ) 1.648 1.641 Z2A is (YþHREE)-rich and more homogeneous than Polyhedral Volume (A˚ 3) 2.27(5) 2.23(4) SWC, with a slight increase in HREE towards the T. Q. E. 1.0090 1.0105 Angle Variance 41.4 48.1 edges of the crystals (Fig. 4e). Electrons at Site 2.78(15) 2.49(10) The EMP analyses were obtained from thin sections of samples SWC and Z2A, and from the T site SWC Z2A grain mount of SWC. Representative average analyses and corresponding atoms per formula unit values in ˚ T–O1 (A) 1.606(5) 1.602(4) high- and low-density regions are given in Table 6. T–O2 (A˚ ) 1.646(5) 1.649(4) The gadolinite formula was first normalized to 10 T–O3 (A˚ ) 1.631(5) 1.618(4) oxygen anions, and the Be content was recalculated by T–O4 (A˚ ) 1.654(5) 1.649(4) ,T–O. (A˚ ) 1.634 1.629 further assuming 4 apfu of Si þ Be and no hingganite Polyhedral Volume (A˚ 3) 2.23(3) 2.21(2) or datolite component (i.e., no hydroxylation of the u T. Q. E. 1.0028 1.0045 site). In contrast, the simpler constraint of 2 apfu of Be Angle Variance 11.6 18.3 appears inaccurate, as both the EMP and the XRD Electrons at Site 8.70(16) 9.40(11) results suggest an excess of Si in some samples (see discussion below). Based on this first normalization, analyses from sample SWC tend to yield an abnor- mally low M-site occupancy, which suggests the presence of vacancies and a substitution towards hingganite or datolite. Since the B content is very low (as determined by LA-ICP-MS, see below), we
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 on 15 April 2021 by University of Colorado Boulder user Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf 836
TABLE 4. ATOMIC POSITION, DISPLACEMENT PARAMETERS, AND ELECTROSTATIC ENERGIES FOR GADOLINITE SAMPLES
Electrostatic SWC xy zU11 U22 U33 U23 U13 U12 Potential (V) A(REE) 0.99856(8) 0.10804(6) 0.32796(4) 0.0139(2) 0.0146(2) 0.0112(2) 0.0012(1) 0.0018(1) –0.0003(2) –29.39 M(Fe) 0.0 0.0 0.0 0.0117(7) 0.0272(9) 0.0099(7) 0.0058(5) –0.0002(4) –0.0007(6) –23.70 Q(Be) 0.5331(12) 0.4136(8) 0.3365(6) 0.014(4) 0.017(7) 0.013(3) 0.001(2) –0.001(2) 0.001(2) –30.94 T (Si) 0.4793(4) 0.2730(2) 0.07892(17) 0.0087(8) 0.0090(9) 0.0049(8) –0.0015(5) 0.0002(5) –0.0004(6) –48.41
O1 0.2424(9) 0.4087(6) 0.0324(4) 0.015(2) 0.020(2) 0.016(2) 0.0024(17) –0.0022(2) –0.0039(18) 25.74 MINERALOGIST CANADIAN THE O2 0.6711(9) 0.2877(6) 0.4533(4) 0.013(2) 0.017(2) 0.013(2) 0.0034(15) 0.0032(14) 0.0022(16) 27.13 O3 0.6826(9) 0.3458(6) 0.1959(4) 0.015(2) 0.024(2) 0.009(2) –0.0052(15) 0.0000(13) –0.0045(17) 26.56 O4 0.3167(10) 0.1035(6) 0.1395(4) 0.018(2) 0.0119(18) 0.017(2) 0.0007(16) 0.0033(15) –0.0052(17) 27.54 / (O5) 0.2016(10) 0.4123(7) 0.3348(4) 0.013(2) 0.025(2) 0.016(2) –0.0014(17) –0.0017(15) 0.0014(18) 22.80
Electrostatic Z2A x/a y/b z/c U11 U22 U33 U23 U13 U12 Potential (V) A(REE) 0.00027(7) 0.10748(5) 0.32856(4) 0.00977(13) 0.01366(17) 0.01250(14) –0.00101(9) 0.00053(9) –0.00017(9) –30.15 M(Fe) 0.0 0.0 0.0 0.0078(4) 0.0224(6) 0.0117(5) 0.0043(4) –0.0011(3) 0.0000(4) –23.85 Q(Be) 0.5346(11) 0.4130(8) 0.3349(5) 0.001(2) 0.016(3) 0.014(2) 0.001(2) –0.0006(16) –0.0002(16) –30.88 T(Si) 0.4801(3) 0.27873(19) 0.07792(14) 0.0064(6) 0.0116(7) 0.0099(6) –0.0006(5) –0.0013(4) –0.0003(4) –48.41 O1 0.2399(7) 0.4129(5) 0.0302(4) 0.0107(15) 0.0168(17) 0.0166(16) 0.0020(15) –0.0013(11) 0.0025(12) 25.89 O2 0.6744(7) 0.2869(5) 0.4518(3) 0.0088(14) 0.0155(16) 0.0131(14) 0.0028(13) 0.0006(10) 0.0010(11) 27.34 O3 0.6892(7) 0.3470(5) 0.1947(3) 0.0087(14) 0.0210(18) 0.0115(14) –0.0041(14) –0.0009(10) –0.0039(12) 26.81 O4 0.3136(7) 0.1070(5) 0.1406(4) 0.0106(14) 0.0118(14) 0.0170(15) 0.0005(14) 0.0000(11) –0.0029(12) 27.69 / (O5) 0.2009(7) 0.4123(5) 0.3325(4) 0.0076(14) 0.0212(18) 0.0160(15) –0.0023(15) –0.0020(11) 0.0001(12) 22.98 Be-Si DISORDER AND REE CRYSTAL CHEMISTRY IN GADOLINITE 837
TABLE 5. REFINED STRUCTURAL OCCUPANCY OF Laser ablation-inductively coupled plasma-mass CATION SITES IN GADOLINITE SAMPLES spectrometry
Structural Site SWC Z2A The LA-ICP-MS measurements of sample WC-1A A (Yb3þ, 2) 46.2 electrons 43.6 electrons are given in Supplementary Table 2. The results reveal M (Fe2þ, 1) 88(1)% 95(1)% a small but significant B content varying from ~250 Q (Be2þ, 2) 90(2)% 94(1)% ppm up to ~690 ppm from LREE-rich to HREE-rich Q (Si4þ, 2) 10(2)% 6(1)% domains; the Be content increases slightly from ~3.5 T (Si4þ, 2) 84(2)% 93(1)% to ~4.0 wt.%. The measured REE content is consistent T (Be2þ, 2) 16(2)% 7(1)% with the EMP data. Uranium and Th yield 40–210 and 370–980 ppm, respectively. Other significant elements revealed by these LA-ICP-MS analyses and not can disregard any significant datolite component. In analyzed by EMP include Mn (1154–1313 ppm), Zn order to estimate the hingganite component and a (480–595 ppm), and 11 other elements below 100 ppm theoretical OH content, a second mineral formula (Pb ~ 70 ppm, Al ~ 43 ppm, Mg ~ 42 ppm, Ti ~ 24 recalculation was iteratively performed assuming a ppm, Rb ~ 15 ppm, K ~ 14 ppm, Sn ~ 10 ppm, Li ~ 8 total of 10 total oxygen anions (including O from OH ppm, and Bi, Sr, Ga ~ 5 ppm). All other elements groups), (Si þ Be) ¼ 4, and M-site vacancy ¼ 1 – (Fe þ analyzed are below 5 ppm or below detection limit Ca) ¼ 2 * H. The iteration was stopped when the total (Supplementary Table 2). recalculated cations converged to a total positive charge of 20.000 (¼ 5 to 7 iterations). Results from this DISCUSSION second normalization are presented in Table 6. Variation of chemical compositions The compositional variations within each sample are illustrated by means of chondrite-normalized plots Gadolinite from the White Cloud pegmatite is (Fig. 5a, b) and atom per formula unit (apfu) plots of distinct in exhibiting a large variation in composition the total LREE versus HREE (Fig. 5c), of the ratios Y/ from LREE- to HREE-rich within a single occurrence, Yb versus Ce/Nd (Fig. 5d), and of the total apfu LREE as well as variable Fe content (Fig. 5, Table 6). A versus FeþCa (Fig. 5e). Sample SWC shows the chondrite-normalized plot of sample Z2A reveals a largest compositional variation and is typically LREE- distinct enrichment in HREEþY with variation in dominant or slightly HREE-dominant, ranging from HREE content and a roughly similar LREE content gadolinite-(Ce) to gadolinite-(Y), whereas sample Z2A (Fig. 5a). In contrast, all analyses of sample SWC is more homogeneous and identified as gadolinite-(Y) show similar normalized LREE and HREE values on (Fig. 5, Table 6). Most of the variation in composition average and a variation of LREE content correlated is seen in the REE content and is explained by a simple with a decrease in HREE (Fig. 5b). All samples show a LREE , HREE exchange (Fig. 5c). The actinide strong Eu anomaly (Fig. 5a, b) explained by feldspar contents of both samples are low, with UO2 below crystallization during intrusion of the granite. While detection limits and ThO2 around 400 ppm in Z2A and sample SWC has variable LREE content, sample Z2A slightly higher (between 0.05 and 0.49 wt.%) in SWC has more consistent LREE and variable HREE content (Table 6; Supplementary Table 1); Pb is close to (Fig. 5c, d). detection limit (,130 ppm) and up to 900 ppm. Most other REE-minerals in the pegmatite incor- Calcium is present as a minor element in all samples. porate mostly LREE or HREE, but not both, such as Totals for several analyses of sample SWC are for instance allanite-(Ce) and thalenite-(Y) (Raschke et somewhat low, around 97.5–99.9%, suggesting either al. 2018). Gadolinite with variable compositions has a low estimate of BeO content or the presence of also been identified in other localities. Gadolinite-(Y) hydroxyl, although O–H stretching modes are not seen from Baveno and Cuasso al Monte in the Italian in the Raman spectra. A few EMP measurements Southern Alps (Pezzotta et al. 1999) shows minor including F indicate no significant F content, with compositional variation, comparable to sample Z2A values at or just above detection limit (,0.01 to 0.02 and to the gadolinite-(Y) from sample SWC (Fig. 5c, wt.%), except for less than 5% of all analyses yielding e). Gadolinite-(Y) from Miyazuma-kyo, Yokkaichi, up to 0.68 wt.% F that correlate with a higher Ca Japan (Miyawaki et al. 1984), and other localities content. Preliminary data including a much larger set (Bacˇ´ık et al. 2014) approaches the composition of of elements reveal the presence of trace Mn (~800 sample Z2A. Gadolinite crystals in Bastnas-type¨ ppm), P (~500 ppm), Mg (~150 ppm), and Al (~130 deposits from the Bergslagen mining region of south- ppm), whereas S, Nb, Ta, Ti, As, Sr, and K were not central Sweden (Holtstam & Andersson 2007) show a detected (, 50–100 ppm). similar variation from gadolinite-(Ce), -(Nd), and -(Y),
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 838 THE CANADIAN MINERALOGIST
FIG. 3. Micro-Raman spectroscopy of thin sections of samples Z2A and SWC. (a, b) Representative Raman spectra of SWC and –1 Z2A located in (c, d). Note the small but significant H2O or OH response between 3000 and 3500 cm clearly visible in the SWC samples and less evident in Z2A. (c, d) Lineshape analysis of tetrahedral Be–O stretch vibrations with spatial variation and correlation between peak position (m), width (FWHM; C), and BSE image. (e, f) Correlated change of peak position and peak width of Be–O stretch vibration; an increase in LREE content leads to peak shifting (red-shift) and broadening in both samples, with a stronger peak broadening in SWC due to its higher LREE content. In sample SWC, poor quality Raman responses (gray dots) are attributed to stronger disorder, e.g., ‘‘S1’’ in (a).
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 Be-Si DISORDER AND REE CRYSTAL CHEMISTRY IN GADOLINITE 839
FIG. 4. High-contrast BSE images of gadolinite crystals in samples (a) WC1-A, (b, c) SWC, and (d, e) Z2A. The variation of grayscale correlates with richness in LREE (bright) or HREE (dark). White lines 1–20 and 21–30 in (a) indicate the position of LA-ICP-MS spot analyses. BSE images (c) and (d) are equivalent to Figure 2e, f and 2g, h, respectively.
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 840 THE CANADIAN MINERALOGIST
TABLE 6. REPRESENTATIVE AVERAGE EMP ANALYSES OF GADOLINITE WITH MINERAL FORMULA RECALCULATION ASSUMING 10 OXYGEN ATOMS AND 4 Si þ Be ¼ 4, M-SITE VACANCY ¼ 1 – (FeþCa), AND H ¼ M-SITE VACANCY
SWC Z2A VLow Y Low Y Med Y* High Y Vhigh Y Low Y Med Y High Y Vhigh Y Points 5 3 7 7 3 5 7 5 4
SiO2 (wt.%) 21.81 21.47 21.81 22.18 22.13 22.25 22.12 22.87 24.62 ThO2 0.37 0.12 0.34 0.29 0.06 0.04 0.02 0.04 0.05 UO2 0.01 , 0.01 0.04 0.08 , 0.01 , 0.01 , 0.01 , 0.01 0.02 FeO 10.17 11.31 11.74 11.88 12.04 12.23 12.39 12.74 13.02 CaO 0.25 0.11 0.21 0.19 0.15 0.09 0.08 0.12 0.19
Y2O3 6.76 9.98 10.88 14.22 19.56 18.70 20.18 24.47 28.50 La2O3 4.29 4.80 1.70 1.61 5.75 0.28 0.22 0.24 0.30 Ce2O3 18.26 16.65 11.98 9.98 11.29 2.33 1.72 1.70 1.93 Pr2O3 2.74 2.16 2.38 1.94 1.24 0.77 0.59 0.52 0.52 Nd2O3 10.40 7.81 10.40 8.46 4.04 6.10 5.06 3.95 3.19 Sm2O3 2.71 1.99 3.12 2.59 0.94 3.92 3.64 2.53 1.56 Eu2O3 0.08 0.08 0.09 0.11 0.15 0.06 0.11 0.11 0.13 Gd2O3 2.02 1.61 2.39 2.13 0.83 4.27 4.26 3.17 2.06 Tb2O3 0.48 0.38 0.57 0.52 0.24 0.97 0.97 0.72 0.51 Dy2O3 2.99 2.84 3.78 3.69 1.69 6.49 6.52 5.10 3.63 Ho2O3 0.50 0.56 0.68 0.74 0.39 1.24 1.25 1.08 0.83 Er2O3 1.88 2.30 2.57 2.85 1.86 4.13 4.21 3.84 3.34 Tm2O3 0.36 0.48 0.51 0.60 0.48 0.70 0.71 0.69 0.65 Yb2O3 2.80 4.07 4.15 5.06 4.87 4.64 4.73 5.04 5.30 Lu2O3 0.31 0.54 0.56 0.74 0.85 0.58 0.58 0.69 0.85 BeO** 8.76 8.94 8.87 9.00 9.34 9.16 9.21 9.42 9.60
H2O** 1.18 0.71 0.45 0.49 0.56 0.45 0.38 0.36 0.50 Total 99.13 98.91 99.21 99.35 98.45 99.39 98.96 99.39 101.29 Si (apfu) 2.031 1.996 2.022 2.024 1.983 2.010 1.998 2.009 2.066 Be 1.969 2.004 1.978 1.976 2.017 1.990 2.002 1.991 1.934 Th 0.008 0.003 0.007 0.006 0.001 0.001 0.001 0.001 0.001 U 0.000 - 0.001 0.002 - - - - 0.000 Fe 0.792 0.879 0.910 0.907 0.903 0.924 0.936 0.936 0.914 Ca 0.024 0.011 0.020 0.019 0.014 0.009 0.007 0.011 0.017 Y 0.335 0.494 0.537 0.691 0.933 0.899 0.970 1.144 1.272 La 0.147 0.165 0.058 0.054 0.190 0.009 0.007 0.008 0.009 Ce 0.622 0.567 0.406 0.333 0.370 0.077 0.057 0.055 0.059 Pr 0.093 0.073 0.081 0.065 0.040 0.025 0.020 0.017 0.016 Nd 0.346 0.259 0.344 0.276 0.129 0.197 0.163 0.124 0.096 Sm 0.087 0.064 0.100 0.082 0.029 0.122 0.113 0.077 0.045 Eu 0.003 0.003 0.003 0.003 0.004 0.002 0.003 0.003 0.004 Gd 0.062 0.050 0.073 0.064 0.025 0.128 0.127 0.092 0.057 Tb 0.015 0.012 0.017 0.016 0.007 0.029 0.029 0.021 0.014 Dy 0.090 0.085 0.113 0.109 0.049 0.189 0.190 0.144 0.098 Ho 0.015 0.017 0.020 0.021 0.011 0.035 0.036 0.030 0.022 Er 0.055 0.067 0.075 0.082 0.052 0.117 0.119 0.106 0.088 Tm 0.010 0.014 0.015 0.017 0.013 0.020 0.020 0.019 0.017 Yb 0.079 0.115 0.117 0.141 0.133 0.128 0.130 0.135 0.136 Lu 0.009 0.015 0.016 0.020 0.023 0.016 0.016 0.018 0.022 H 0.367 0.220 0.139 0.149 0.167 0.134 0.114 0.107 0.139 Total 7.160 7.112 7.052 7.056 7.094 7.060 7.058 7.047 7.025
* Analysis ‘‘Med Y’’ of SWC is representative of the crystal chip used for XRD analysis.
** BeO and H2O are estimated via the mineral formula recalculation (see text for detail).
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 Be-Si DISORDER AND REE CRYSTAL CHEMISTRY IN GADOLINITE 841
FIG. 5. Plots of EMP data for sample SWC (orange-red tones) and Z2A (blue tones). (a, b) Chondrite-normalized plots of major gadolinite zones containing similar chemical distributions (chondrite values from McDonough & Sun 1995). (c, d, e) Plots of EMP analyses highlighting variation of (c) total HREE versus total LREE (apfu ¼ atom per formula unit), (d) Y/Yb versus Ce/Nd, and (e) Fe þ Ca versus LREE (apfu). (f) BSE image of the crystal chip of SWC analyzed by XRD showing minimal variations of composition with a few small bright spots corresponding to a LREE-richer domain. A representative averaged composition for this grain is right at the transition from gadolinite-(Ce) to -(Y) (¼ ‘‘SWC Med Y’’ in Table 6). B14: Bacˇ´ık et al. (2014); M84: Miyawaki et al. (1984); P99: Pezzotta et al. (1999); SL78: Segalstad & Larsen (1978); S18: Skodaˇ et al. (2018); HA07: Holtstam & Andersson (2007).
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 842 THE CANADIAN MINERALOGIST
as observed in sample SWC, yet with a higher Y/Yb and u (Table 4). For this calculation, the atoms at the ratio (Fig. 5d). Gadolinite crystals from Skien in the split A-site were ignored. These potentials are useful Norwegian southwestern Oslo region (Segalstad & for predictions of possible heterovalent substitutions as Larsen 1978) have strong core-to-rim zonation, yet are well as stable isotope fractionation. The Y site has a still rich in LREE (Fig. 5c). Although an extensive deep potential that varies inversely with site radius and discussion on this compositional variation is beyond is typical of REE and Y sites in minerals (Smyth & the scope of this paper, the observed change in Bish 1988). The T site is typical of Si tetrahedral sites. composition is likely related to the very late The Q site has a deeper potential than typical Be fractionation of the residual pegmatitic fluid from tetrahedral sites that range from 28.9V for bromellite LREE-rich to HREE-rich (Fig. 4). to 30.3V for phenacite M2. The M site is larger and The EMP analysis of samples SWC and Z2A has a shallower potential than typical Fe2þ sites. The yielded 0.788–0.931 and 0.900–0.951 apfu Fe, respec- oxygen site potentials are also typical of non-bridging tively (Table 6; Supplementary Table 1). The XRD oxygen sites in silicate minerals, and the non-silicate data support the relative amount of Fe in the two oxygen (u) has the shallowest potential as do non- samples: at the M site, SWC has 0.87 apfu Fe and Z2A silicate oxygen positions in other silicate minerals has 0.95 apfu (Table 5). These results suggest the (Smyth & Bish 1988). presence of vacancies at the M site, especially in sample SWC. As B analysis by LA-ICP-MS yielded Low hydroxyl content at the u site mostly less than 700 ppm (Supplementary Table 2), In the absence of iron and the presence of hydroxyl the datolite component can be ignored. Similarly, the at the u site, a solid solution with hingganite or herderite component is also ignored, as only ~500 datolite should form via the exchange M(Fe,Ca)þ2 ppm P was detected by EMP (Supplementary Table 1). þ u2O–2 MA u2(F,OH)–1. Indeed, Raman spectra No significant Li was detected by LA-ICP-MS (~8 þ indicate the presence of a small amount of water or ppm). Therefore, we suggest that the calculated other OH species for both SWC and Z2A (Fig. 3). vacancies at the M site chiefly represent a substitution Similarly, Ca´mara et al. (2008) reported 0.29 apfu towards hingganite, with a double hydroxylation of the OH–1 at the u site in gadolinite-(Y), and Skodaˇ et al. u-site compensating for the vacancy left by the (2018) documented a small OH–1 peak at 3525 cm–1 in divalent cation (e.g., Demartin et al. 2001, Holtstam gadolinite-(Nd) with 0.337 apfu OH–1 in their & Andersson 2007). The decrease in HREE and specimen. increase in LREE at the A site correlates with lower 2 We cannot entirely rule out the presence of minor Fe þ at the M site (Table 6, Fig. 5e), which is in F–1 instead of OH–1 at the u site. Due to the low count agreement with Bacˇ´ık et al. (2014). This observation rate on the EMP (low X-ray yield, strong absorption) suggests a preference for hydroxylation at higher and a strong overlap of F Ka with Ce Mf, the detection LREE-content in gadolinite (Fig. 5e). The total REE limit for F is high. Additional EMP analyses including cations at the A site is close to 2 apfu in most analyses F mainly yield values near the detection limit (,0.01 (1.99 6 0.03). to 0.02 wt.%), and the rare abnormal high F-contents The cell parameters are among the largest reported up to 0.68 wt.% F correlates with higher Ca values, for gadolinite minerals, with values similar to which is likely due to either small fluorite inclusions or gadolinite-(Nd) (Skodaˇ et al. 2018, Bacˇ´ık et al. a secondary X-ray effect from the abundant surround- 2017). The A-M polyhedral layer expands to accom- ing fluorite. Fluorine is certainly abundant in the rock, modate the LREE. This expansion correlates with the yet it appears unlikely that a significant quantity of F M-site vacancy (Fig. 5e). This is seen most promi- substitutes at the u site in the investigated samples. nently in expansion of the A polyhedron, from 24.74 The mineral formula recalculation suggests that A˚ 3 in Z2A to 26.26 A˚ 3, in SWC (Table 3a, b). Unlike possibly up to 0.11–0.14 apfu OH is present in sample the A-M layer, the Q-T layer exhibits only minor Z2A, and values up to 0.37 apfu in SWC (Table 6; changes in polyhedral volumes as the larger A and M Supplementary Table 1). These values are only cation are accommodated by flexure of the tetrahedral approximate, likely represent a maximum content, rings of Si and Be (Fig. 1). and are qualitatively consistent with the Raman Electrostatic site potentials observations. The accuracy of the recalculated water content depends heavily on the quality of the data and Electrostatic energy potentials were calculated the correctness of all assumptions, including the using the code ELEN (Smyth 1989) based on a oxidation state of each element (especially Fe) that nominal point charge model with charges of þ3 for A, can affect the charge balance. Nonetheless, they þ2 for M and Q, þ4 for T, and –2 for O1, O2, O3, O4, should not be too far from the truth, as all EPMA
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 Be-Si DISORDER AND REE CRYSTAL CHEMISTRY IN GADOLINITE 843
analysis totals are between 98.5 and 101.5 wt.% in common rock-forming silicates, particularly layer (including the recalculated H2O and BeO content), and and framework silicates, may be widely overlooked. no additional elements were found in significant quantity by LA-ICP-MS, although minor amounts of CONCLUSIONS Mn, Zn, and B were identified (Supplementary Table Gadolinite from the White Cloud pegmatite is 2). structurally flexible and chemically variable, ranging Be–Si disorder from gadolinite-(Ce) to gadolinite-(Y) dominant members. The high variability in the LREE versus Sample SWC is the first documented case of HREE contents between two distinct gadolinite significant Be–Si disorder in a mineral from XRD samples and the observed chemical zonation within structural refinement. This is a robust conclusion each gadolinite crystal reflect changing fluid or silica- because of the large difference in X-ray scattering rich liquid composition or other environmental between Be and Si. Demartin et al. (1993) suspected parameters during REE crystallization. These speci- that this was a possibility and considered it in their mens are also notable in that they have significant M- analyses of gadolinite from various fissure and granitic site vacancies, which are higher in the most LREE-rich pegmatite localities in the Alps but found no evidence domains, and very limited B. These vacancies are of Be–Si disorder or substitution in their XRD data. explained by the hydroxylation of the u site with no Demartin et al. (2001) addressed Be–Si disorder, but evidence for a significant amount of F substituting for due to the lack of convincing evidence at the time, OH. Thus, there is a small but significant solid-state used the absence of evidence as an assumption for solution with hingganite, with very minimal compo- their chemical formula calculations. Skodaˇ et al. nent from datolite or other endmembers of the (2016) did observe minor Be–Si disorder (under 5%) gadolinite supergroup. in their analysis of gadolinite-(Nd), but they did not Although beyond the scope of this paper, the elaborate on this topic. The current data offer a robust enrichment of REE in the liquid could be explained indication of significant Be-Si disorder between through the process of constitutional zone refining crystallographically distinct tetrahedral sites. (e.g., London 2018), and the observed mineral The Q site of sample SWC has 88(2)% Be and chemistry could be the result of a final fractional 12(2)% Si, while the T site has 86(2)% Si and 15(2)% crystallization of the residual liquid in the core of the Be. The Q site of Z2A has 93(1)% Be and 7(1)% Si, pegmatite with an initial supersaturation of LREE and and the T site has 94(1)% Si and 6(1)% Be (Table 5). enrichment of the remaining liquid in HREE at The disordering of Si, and potential to have excess Si undercooling conditions (London 2014). The mineral- at the Q site, or deficient Si at the T site, indicates that ogy suggests that the mineralizing liquid was Si- and one cannot assume 2 apfu for Si a priori when F-rich with limited Be, yet more work is required to normalizing the chemical formula of gadolinite. constrain its exact composition. However, a total of four Be þ Si per formula unit is The disorder of Be at the T site provides direct consistent with the XRD data for gadolinite. There is evidence that Be can substitute into Si tetrahedra. no evidence for vacancies at either the Q or T sites. This Be–Si disorder could be present in more than The occupants are chiefly Si or Be, as LA-ICP-MS only gadolinite minerals, due to the omnipresence of Si in tetrahedral coordination in minerals of the crust revealed less than 800 ppm B and ~8 ppm Li and upper mantle. If Be is able to substitute for Si, or (Supplementary Table 2), which is consistent with even Al, it could be present in many rock-forming the observation that the White Cloud Pegmatite and minerals as a trace element. Disorder could be found Pikes Peak Granite in general are depleted in B in other Be-silicates, such as phenakite, beryl, and (Simmons & Heinrich 1980). euclase, and these are strong candidates for seeking The Q and T site tetrahedra are very similar in Be–Si disorder in future studies. Since Be (or B) is volume and bond lengths, so they are fit for ionic not regularly analyzed for, its overall abundance and substitutions (Table 3a, b). The Q sites of the samples geochemical significance may not yet have been fully have tetrahedral volumes of 2.27 A˚ 3 and 2.23 A˚ 3 for recognized. SWC and Z2A, respectively. The T sites of samples ˚ 3 ˚ 3 SWC and Z2A have volumes of 2.23 A and 2.21 A , ACKNOWLEDGMENTS respectively, which is minimally less than that of the Q site. The Be tetrahedra are relatively undistorted, This study was funded in part by grant NSF-EAR similar to the Si tetrahedra. Beryllium is a relatively 1416979 to JRS, by the Bolyard Scholarship from the rare element and not commonly analyzed due to Rocky Mountain Association of Geologists to REH, analytical difficulty for light elements, so its presence and by a student fellowship from Mile High Rock and
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 844 THE CANADIAN MINERALOGIST
Mineral Society to REH. JMA acknowledges funding GUILLONG, M., VON QUADT, A., SAKATA, S., PEYTCHEVA, I., & from NSF EAR-1427626 and the University of BACHMANN, O. (2014) LA-ICP-MS Pb-U dating of young Colorado Boulder for purchase of the JEOL JXA- zircons from the Kos-Nisyros volcanic centre, SE Aegean arc. Journal of Analytical Atomic Spectrometry 29, 963– 8230 electron microprobe. We thank Dr. M. Guillong 970. for providing the LA-ICP-MS analyses. We greatly appreciate comments and suggestions from the two HAYNES, C.V. (1965) Genesis of the White Cloud and related reviewers, Dr. R. Skodaˇ and Dr. W. Simmons, and pegmatites, South Platte Area, Jefferson County, Colo- from Guest Editor T. Martins. rado. Geological Society of America Bulletin 76, 441– 462. REFERENCES HOLTSTAM,D.&ANDERSSON, U.B. (2007) The REE minerals of the Bastnas-type deposits, South-Central Sweden. ALLAZ, J.M., WILLIAMS, M.L., JERCINOVIC, M.J., GOEMANN, K., Canadian Mineralogist 45, 1073–1114. &DONOVAN, J. (2019) Multipoint background analysis: Gaining precision and accuracy in microprobe trace LONDON, D. (2014) Subsolidus isothermal fractional crystal- element analysis. Microscopy and Microanalysis 25, lization. American Mineralogist 99, 543–546. 30–46. LONDON, D. (2018) Ore-forming processes within granitic AZAVANT,P.&LICHANOT, A (1993) X-ray scattering factors of pegmatites. Ore Geology Reviews 101, 349–383. oxygen and sulfur ions: An ab initio Hartree-Fock calculation. Acta Crystallographica (A)49, 91–97. MCDONOUGH, W.F. & SUN, S.-S. (1995) The composition of the Earth. Chemical Geology 120, 223–253. BACˇ I´K, P., FRIDRICHOVA´ , J., UHER, P., PRSˇEK, J., & ONDREJKA, M. (2014) Crystal chemistry of gadolinite-datolite group MIYAWAKI,R.,NAKAI,I.,&NAGASHIMA, K. (1984) A silicates. Canadian Mineralogist 50, 1–17. refinement of the crystal structure of gadolinite. American Mineralogist 69, 948–953. BACˇ I´K, P., MIYAWAKI, R., ATENCIO,D.,CA´ MARA,F.,& FRIDRICHOVA´ , J. (2017) Nomenclature of the gadolinite NILSSEN, B. (1973) Gadolinite from Hundholmen, Tysfjord, supergroup. European Journal of Mineralogy 29, 1–16. North Norway. Contribution to the Mineralogy of Norway, No. 55. Norsk Geologisk Tidskrift 53, 343–348. CA´ MARA, F., OBERTI, R., OTTOLINI, L., DELLA VENTURA, G., & BELLATRECCIA, F. (2008) The crystal chemistry of Li in PEZZOTTA, F., DIELLA, V., & GUASTONI, A. (1999) Chemical gadolinite. American Mineralogist 93, 998–1004. and paragenetic data on gadolinite-group minerals from Baveno and Cuasso al Monte, southern Alps, Italy. Cˇ ERNY´ ,P.&ERCIT, T.S. (2005) The classification of granitic American Mineralogist 84, 782–789. pegmatites revisited. Canadian Mineralogist 43(6), 2005– 2026. RASCHKE, M.B., ANDERSON, E.J.D., VANFOSSON, J., ALLAZ, J.M., SMYTH, J.R., Sˇ KODA, R., PERSSON, P.M., & BECKER,R. CROMER, D.T. & MANN, J. (1968) X-ray scattering factors (2018) Rare-earth crystal chemistry of thalenite´ from computed from numerical Hartree-Fock wave functions. different environments. Mineralogical Magazine 82, 313– Acta Crystallographica (A)24, 321–325. 327.
DEMARTIN,F.,PILATI,T.,DIELLA,V.,GENTILE,P.,& SEGALSTAD, T.V. & LARSEN, A.O. (1978) Gadolinite-(Ce) from GRAMACCIOLI, C.M. (1993) A crystal-chemical investiga- Skien, southwestern Oslo region, Norway. American tion of Alpine gadolinite. Canadian Mineralogist 31, Mineralogist 63, 188–195. 127–136. SHELDRICK, G.M. (2015) Crystal structure refinement with DEMARTIN, F., MINAGLIA, A., & GRAMACCIOLI, C.M. (2001) SHELXL. Acta Crystallographica (C)71, 3–8. Characterization of gadolinite-group minerals using crystallographic data only: The case of hingganite-(Y) SIMMONS, W.B., JR.&HEINRICH, E.W. (1980) Rare earth from Cuasso al Monte, Italy. Canadian Mineralogist 39, pegmatites of the South Platte District, Colorado. 1105–1114. Colorado Geological Survey Research Series 11, 131.
DONOVAN, J.J. & TINGLE, T.N. (1996) An improved mean Sˇ KODA, R., PLA´ SIL, J., Cˇ OPJAKOVA´ , R., NOVA´ K, M., JONSSON, E., atomic number correction for quantitative microanalysis. GALIOVA´ , V.M., & HOLTSTAM, D. (2016) Gadolinite-(Nd), Journal of Microscopy 2, 1–7. IMA 2016-013. Mineralogical Magazine 80, 916.
FARRUGIA, L.J. (2014) WinGX – Version 2014.1. An Sˇ KODA, R., PLA´ SˇIL, J., Cˇ OPJAKOVA´ , R., NOVA´ K, M., JONSSON, E., integrated system of windows programs for the solution, GALIOVA´ , M.V., & HOLTSTAM, D. (2018) Gadolinite-(Nd), refinement and analysis of single crystal X-ray diffraction a new member of the gadolinite supergroup from Fe-REE data. Journal of Applied Crystallography 45, 849–854. deposits of Bastnas-type,¨ Sweden. Mineralogical Maga- zine 82, 133–145. GIBSON, S.J. & EHLMANN, A.J. (1970) Annealing characteris- tics of metamict gadolinite from Rode Ranch Texas. SMITH, D.R., NOBLETT, J., WOBUS, R.A., UNRUH, D., DOUGLASS, American Mineralogist 55, 288–291. J., BEANE,R.,DAVIS,C.,GOLDMAN,S.,KAY,G.,
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021 Be-Si DISORDER AND REE CRYSTAL CHEMISTRY IN GADOLINITE 845
GUSTAVSON, B., SALTOUN, B., & STEWART, J. (1999) SMYTH, J.R. & BISH, D.L. (1988) Crystal Structures and Petrology and geochemistry of late-stage intrusions of Cation Sites of the Rock-Forming Minerals. Allen & the A-type, mid-Proterozoic Pikes Peak Batholith (Central Colorado, USA): Implications for petrogenetic models. Unwin, Boston, Massachusetts, United States. Precambrian Research 98, 271–305.
SMYTH, J.R. (1989) Electrostatic characterization of oxygen sites in minerals. Geochimica et Cosmochimica Acta 53, Received October 26, 2019. Revised manuscript accepted 1101–1110. March 6, 2020.
Downloaded from http://pubs.geoscienceworld.org/canmin/article-pdf/58/6/829/5275560/i1499-1276-58-6-829.pdf by University of Colorado Boulder user on 15 April 2021